Aiming to become the global leader in chip-scale photonic solutions by deploying Optical Interposer technology to enable the seamless integration of electronics and photonics for a broad range of vertical market applications

Free
Message: Sidoti Transcript

Here is my version of the Sidoti zoom call. Errors are mine (and Mika's).

Rogue

 

Sidoti - Poet transcript

Hi. Good afternoon. My name is Tom Mika and I am the executive vice president and CFO of Poet Technologies. In addition to those two roles I am kind of the explainer in chief. So, I have called this my basics presentation. Because I can’t possibly explain in the thirty minutes that I have available the technology that has been developed in the company. And with all due respect to rocket scientists out there, this stuff is way more complex that rocket science. So if you have any questions about what I am presenting or if you have any questions about the technology please contact me and I am happy to connect you with people who can really explain the technology in depth. I’ve had some great one-on-ones this morning including with people who’ve been invested in other photonics companies. So they were great one-on-ones.

I can’t really go very far in this presentation without pointing out, you know, Suresh Venkatesan, our chief executive officer, who’s behind the development of this technology. He’s brought in a great team. And particularly in the last year he brought in Vivek Rajgharia who headed up the photonics division at Macom. Vivek has brought in some other great people on the photonics side of the business. So we have this great balance of depth of knowledge on semiconductor integration and the business side of photonics within the company. So I am going to go ahead with this presentation, and sometimes I do a good job and sometimes I don’t. And my shareholders always tell me when I do and when I don’t so we’ll see how this goes.

[safe harbor info…]

[2:36] So we are a company of about 30 employees. We’ve got a very deep patent portfolio. Out of those 16 pending patents, includes 3 what we call provisional patents that usually end up being multiples of what they start out to be with the claims. We have two basic technology offerings. Most of you... many of you on this call might be familiar with our Poet optical interposer platform and the designs that we’ve done for lasers and detectors that interact with our optical interposer.

From a company point of view, we are headquartered in Canada and we have a few operating divisions. I don’t usually talk about ODIS. That’s the North American... its a Delaware corporation, but now that we have a locus in Allentown, Pennsylvania where Vivek and his team reside, I thought I’d point that one out. We also have a location in Singapore and we’re just opening a location in Shenzhen in China. The reason that we are in these places is that this is where the talent in photonics resides. We are always looking for talent, and so you have to be where the talent is. We also have one joint venture, Super Photonics Xiamen of which I will go into some detail later.

[4:18] So I can’t stay away from all of the technology. We are a photonics company, and so I wanted to get at least a basic understanding about what photonics is. Photonics has to do with the creation, detection, and manipulation of light. Particularly laser light that is really very fundamental to computing and telecommunications. And some of the biggest trends, growth trends in computing today including those that are mentioned on this page.

Virtually all data communications, for example, within mega data centers is handled on fiber optics. And any time there is a fiber optic connection, that is a potential market for Poet. Cloud computing as everyone knows is growing rapidly. Data centers… Mega data centers are actually being built at about the rate of about one per month by all these data center operators in the world. Artificial Intelligence is a service now available in the cloud, soon to be available in what we call edge computing. And we are engaged with a company involved with neuromorphic optical computing which I’ll go into a little bit later. And the whole trend of 5G and IOT is to push more computing power down to the edge of the network. Facial recognition for example is something that is being pushed to the edge. And that is where neuromorphic optical computing actually is a good example of an edge computing activity.

[6:12] So what’s the problem you think? Its just a collection of electronic and light devices, and lasers and detectors and so on. What’s the big deal? Well the big deal is that making these devices that are reliable is very expensive in both capital and labor. And because its not all semiconductors, it hasn’t kept up with Moore’s law. So, even today, most photonics devices are built one at a time with armies of operators behind these warehouses full of testing machines. The technology that is being used today requires what we call the active testing. So anytime you put one of these devices together, say for example a laser and lens… a laser and another device, a waveguide… It has to be tested. It has to be optimized and that’s a very expensive and capital intensive process.

[7:17] So this has been recognized for a long time. The integration of components at wafer scale is something that many large companies, one in particular, Intel, has pursued for the past 20 years. And they’ve not been successful at going full wafer scale. No one has. Until now.

[7:46] Poet, and Suresh in particular, took on a dual challenge of both being an integration device and a platform. To develop a unique and disruptive and differentiated entry into the photonics market. His philosophy was, and is, ”if you cant bring something to the market that is sustainable and has a sustainable differentiation, why bother?”

Back in late 2015, early 2016 we acquired two companies. One was a fab in Singapore, Denselight Semiconductors. Another was one called BB Photonics. And he was actually prescient in understanding that a problem that he had worked on for several years prior to joining Poet, which was called silicon photonics… and the implementation of silicon photonics, that the BB Photonics technology could be important to producing something that was truly differentiating in photonics. And it turned out to be.

[9:07] But rather than just building one device that could solve a problem in cost or integration, what he took on was the challenge of building platforms. Meaning that he could and we could, as Poet, bring together a group of technologies built on a single platform and use that as a base for other applications. And a very good example, probably the best example in the world of this kind of dual integration and platform technology is of course the smart phone, which both integrates a lot of different applications and is a great platform for other apps for people to develop it.

[10:00] The Poet optical interposer is that kind of invention. Its patented and enables lower cost and higher performance in photonics devices across a broad range of applications. The schematic shown here may not be meaningful, but let me take you through a little bit of explanation. First of all we are creating this device on a standard silicon wafer, which is different than the other companies today trying to do this. They use SOI which is higher cost.

[10:39] Being a standard silicon wafer it has metal traces that run through the wafer. So those metal traces are used for the intercommunication of devices. What’s brilliant about this design is that all of the devices that are incorporated into the interposer are on the top layer. What we do is take a silicon wafer and deposit a dielectric layer and etch some of that dielectric off to create what we call waveguides. And waveguides are to light what a printed circuit is to electronics. An electron will follow a copper wire.. in a copper trace on a printed circuit board. But light won’t. Light has to be guided in a special way because it is a wave. That waveguide material is very unique and its proprietary to Poet, and is foundational to the technology that we have. It allows us to do what we do, which is, to create these devices at wafer scale, which means that we’ve build them 100’s at a time rather than 1 at time. It also allows high speed communication between the optical devices and the electrical devices that are on the same interposer. So the architecture of the device is really fundamental to it and fundamental to its flexibility and our ability to use it for different applications.

[12:28] So what’s the benefit? Well there are some terrific benefits to the whole optical interposer approach. Its dramatically lower cost. So in some applications in some devices, we can produce a device that does the same functions but for 25% to 40% lower cost, depending on the application. This is a particularly important aspect of the optical interposer in a market that we are focused on initially which is the optical transceiver market where cost is everything.

[13:11] What some people... sometimes I don’t get to the point of being able to also explain that for the same volume of devices our capital investment making those devices is only about 10% of the capital which is required to produce the same number of devices by conventional means, or even by the 12” silicon based vendors.

[13:40] We are at chip scale which means that compared to conventional… some others… we’re tiny. At 50% to 90% reduction in the real estate. So we’re 10% to 50% of the actual size on the silicon wafer. Which means that we can put more on a wafer. We do everything at wafer level assembly and test because that’s really fundamental to getting the cost advantages that we have. If we can’t do it at wafer level, we’re not really interested in doing it.

[14:15] I explained before that the planar architecture allows the ease of production. The production of it is quite simple to understand. We have a wafer that contains 100’s of optical interposers. That wafer goes into an assembly line in which devices which are lenses in the case of optics or lasers in the case of photonics... detectors... are picked up by a machine and placed down on the surface of the wafer. There is no requirement for testing that connection. And we don’t have to test that connection because we know that placement will be in a tolerance that we can afford because of the waveguide material that we use.

[15:08] And finally as a platform technology we can build the device once. We don’t have to build it ten times for ten different kinds of applications or the next generation of a particular device. So, in prior presentations I have said that we’ve built an interposer that is a tiny device, 6.5mm by 10mm. It can be used... the same device, the same multiplexer, the same waveguides are used for 100G speeds of transceiver devices, for 200G, and for 400G.

The proprietary waveguides that are produced in the Poet optical interposer are done in a foundry in Malaysia. The company’s name is Silterra Malaysia. That technology is owned by us. We own the process technology. We own the equipment on which key aspects of the technology are done. And that has been under tight containment for a very long time. At the current time we have no intention to license that fundamental technology. And we have not licensed that technology in connection with the joint venture that we have entered.

[16:40] So we’ve articulated our business and our vision and mission as taking an industry leadership position in integrated photonics. Because we believe that this is truly a disruptive platform device. And as I said before, anything where you have fiber, and that includes data centers, it includes 5G networks, and if it is self-contained, in light, in the case of the neuromorphic optical engine, we can play a role in that. Our markets, our available markets, are pretty vast.

[17:24] Our short term business plan is to make prototypes of these devices that we call optical engines which contain all of the elements of a device that deal with the light. For several applications. And we are dealing with those applications with potential customers, and we are going to be providing those as samples to customers within the next few months.

Right now what we are doing is we’re taking data from our optical interposer based prototypes and providing that data to a long list of customers that we’ve engaged with over the past two years… primarily in PowerPoints, and all but a few of those companies have said to us “We understand what you are doing. We think its great. Come back to us when you have some data and you have a device that you can share with us.” So we’re now at that point of being able to do that… to provide data on the performance of devices which is actually exceeding our own internal expectations for how well they are performing. And we are going to be sampling our prototypes for qualification.

[18:50] What happens in this business is you’ve demonstrated a technology. If the customer is interested you enter into a design-in phase. And that design phase will take 3 to 6 months of adapting our optical interposer for their particular application. And then providing prototypes which are evaluated. Typically they are what we call beta prototypes. And those prototypes then go into qualification. No customer wants field failures of their devices once they are out in the field. So it’s a very rigorous qualification process that operates for… depending on the company or depending on the application, for 3 or 6 or 9 months. Once they are qualified they go into full production. So we are right at the precipice of being able to go back to customers and prove that we can do what we told them, over the last two years, that we could do. And it’s a very long list of prominent companies in North America and in China.

[20:10]

 [NOTE: list of products on slide: 100G/200G optical engines, 400G Light Bar, 400g FR4 receiver, 400G FR4/DR optical engines, C-Band DWDM LighttBar, 200G LR4 optical engines]

All of the products, at least the top 4 products on this list, we’re doing in connection with a particular customer. And I won’t go through what all of these do. But what I will say about them is they are all related to transceivers and the optical transceiver market.

[20:37] The 400G Light Bar, that has a trademark associated with it, is also a product that we think is applicable to the co-packaged optics market. And briefly I’ll just say that what’s happening in that market is that Broadcom’s latest switch that operates at 25.6 Tb/sec. Once it goes to the next generation that pretty much doubles that speed… Much of the energy that is available to power the switch blade and the server racks is going to be used up. Something has to happen. And what has to happen is all of the compensating circuits that are currently in a switch blade that compensate for the loss of power and the timing of the signal within that box when the electrons go from the switch to the back side of the blade… to the backplane… All of those circuits that also take energy have to come out. And we think we are in a very good position to offer products in that area because of the technology that we’ve got in the optical interposer. We think we can come much closer to the switch itself and provide some flexibility and lower power requirement solutions in the market.

[22:30] One of the applications that we are working on with a customer is in this optical neuromorphic processing area. It turns out that using optics is preferable for some kinds of matrix algebra that’s used in artificial intelligence, particularly in inference engines, which develop recommendations or come to conclusions about behavior or about data. If you can move that processing to a chip and put it at the edge of the network, then what you can avoid is a lot of communication from your cameras, for example, in a facial recognition system. That’s taking in a lot of data. All of that data now is moving to the cloud. Its being processed in the artificial intelligence systems up there in the cloud. Then it comes back with recommendations. Well, if you can move that to the edge and put it in the camera itself, then you can avoid all of that expense of going back. If you think about it that applies to everything in the IOT. Anything where you need to develop recommendations or you need to pattern behavior or you need to infer based on what is actually happening in the field.

[24:00] So I said we are very close and therefore we have customers who are imminent. What I am showing here is that we have to take our chip which is this sort of red and purple thing here and put it into something so that it communicates with our instrumentation and our customers instrumentation in order to do a live demo. And that is what we are engaged in doing now. We can do that with alpha prototypes. So, we are right at the point where we are getting alpha prototypes. We are taking data. We will be providing data. But we will also be here very soon doing live demos in either the customer lab, if covid allows it, or we can do it live in our own lab. Of course, we are planning being at OFC and we will be doing live demos at OFC, which is a big optical conference which will be in San Francisco this year in June.

[25:00] The markets here are large. What I am representing here are the SAMs this is where we believe we can directly compete. Its not the total available market. The total available market is probably 2 to 3 times bigger. But these markets while they are large are absolutely dwarfed by two others that I mentioned. One was the co-packaged optics market and the other is the biggest of them all, is the edge computing market, which is just beginning and where we believe we can provide a significant advantage.

[25:40]

[NOTE – slide lists company names: in areas of transceiver modules, optical systems (Cisco, Huawei, Juniper, Acacia, Arista, Nokia) and cloud data center companies (AWS, Google, Facebook, AliBaba]

Now many people ask me “who are you going to sell this to?”, particularly the optical interposer for optical transceivers. Well because of the price advantage that we can provide and the performance advantages that we can provide, we assemble a list of all the major transceiver module producers. One thing here is that we are not ourselves going to be competing with them with our own module. What we are doing is we’re replacing some of the components within them with our optical engines that cost 25% to 40% less than those components do and offering these transceiver module makers the ability to gain some profitability. Where right now because of the pressure that the cloud data centers put on these transceiver module manufacturers they’re not making any margin at all. But the cost savings and the architecture of the optical interposer is so differentiating that the cloud data center customers are also our customer base. Because they often dictate to others in the value chain what should be used in their data center. So we’re intending and have had discussions with companies under the cloud data center list.

[27:15] You can show a customer your technology the very next question they ask after they say “its great technology” is “How are you going to delivery 100,000s or millions of units to us?” And Super Photonics Xiamen is our answer to that.

Sanan optoelectronics produces 25M 6” wafers a year. They are the largest compound semiconductor manufacturer in the world. Our joint venture with them is signed back on October 22. We’re putting in our assembly and test technology. We will also be selling to the joint venture our optical interposer platform. Sanan will be providing selling to the joint venture lasers and detectors. Sanan puts up 25M in cash to buy equipment and facilities and cover all operating expenses of the joint venture. And we put in the IP and no cash. We believe that is a very good move for us and it solves all of the questions for customers about how are you going to manufacture.

[28:30] So our manufacturing strategy is outlined here. It is really a consolidation supply chain and we intend to source our photonic devices from our former subsidiary Denselight, Almae Technologies in France and SAIC in China.

[28:50] In summary, we spent two or three years on technology development. Since about mid-2017 all along really knowing that ultimately what we wanted was the optical interposer and that we needed devices that we could… a design that would… could be used on the optical interposer. In 2019 coming into 2020, we completed proof of concept with a tier 1 customer. Still engaged with that customer. And two additional customers in design projects. Both of which include supply agreements. We’re providing the data packages to a long list of customers. We are currently staffing and acquiring equipment for the Super Photonics in Xiamen. And frankly I am glad we have an audience today because we are hardly known in the United States. And we are covered by Kevin Dede at Wainwright and Lisa Thompson at Zack’s. I am happy to take questions. [NO TIME WAS LEFT FOR QUESTIONS]

Share
New Message
Please login to post a reply